c50py 0.2.2__tar.gz → 0.2.4__tar.gz

{c50py-0.2.2/src/c50py.egg-info → c50py-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: c50py
-Version: 0.2.2
+Version: 0.2.4
 Summary: C5.0-like Decision Trees in Python (scikit-learn style)
 Author-email: David Díaz <daviddiazsolis@gmail.com>
 License: MIT
@@ -18,10 +18,11 @@ Requires-Dist: numpy>=1.21
 Requires-Dist: scikit-learn>=1.2
 Provides-Extra: graphviz
 Requires-Dist: graphviz>=0.20; extra == "graphviz"
+Dynamic: license-file
 
-# c5py — C5.0‑style Decision Trees for Python (clean 0.2.0)
+# c50py — C5.0‑style Decision Trees for Python (clean 0.2.0)
 
-`c5py` provides transparent, easily inspectable decision trees modelled on
+`c50py` provides transparent, easily inspectable decision trees modelled on
 Quinlan’s C5.0 algorithm.  Both classification and regression trees are
 supported and expose a scikit‑learn‑like API.  The implementation is written
 from scratch in pure Python/Numpy and includes support for numeric and
@@ -86,7 +87,7 @@ Fit a regression tree to the diabetes dataset and obtain a visualisation:
 ```python
 import pandas as pd
 from time import perf_counter
-from c5py import C5Regressor
+from c50py import C5Regressor
 
 df = pd.read_csv("diabetes.csv")
 y = df["target"].values

{c50py-0.2.2 → c50py-0.2.4}/README.md

@@ -1,114 +1,114 @@
-# c5py — C5.0‑style Decision Trees for Python (clean 0.2.0)
-
-`c5py` provides transparent, easily inspectable decision trees modelled on
-Quinlan’s C5.0 algorithm.  Both classification and regression trees are
-supported and expose a scikit‑learn‑like API.  The implementation is written
-from scratch in pure Python/Numpy and includes support for numeric and
-categorical variables, missing values, pre‑ and post‑pruning, boosting,
-rule tracing/export and Graphviz visualisation.
-
-## Features
-
-- **Scikit-learn API:** `fit(X, y)`, `predict(X)`, `score(X, y)`.
-- **Categorical support:** Pass `categorical_features=[0, 2]` to handle categories natively.
-- **Sample weights:** Supports `sample_weight` in `fit` for weighted splitting and pruning.
-- **Missing values:** Handles missing values using C5.0's fractional propagation strategy.
-- **Boosting:** Set `trials=10` to train a boosted ensemble.
-- **Rule export:** call `export_rules()` to get a list of human-readable rules.
-- **Graphviz export:** call `export_graphviz()` to visualize the tree.
-- **Pretty printing:** call `print_tree` to display the learned splits in a
-  readable nested `if`/`else` format (single trees only).
-
-## Documentation
-
-For a comprehensive guide on how to use `c50py`, including advanced features and examples, please see the [Usage Guide](USAGE_GUIDE.md).
-
-## Installation (development mode)
-
-Install the package into your environment in editable mode:
-
-```bash
-pip install -e .
-```
-
-## Quickstart (Classification)
-
-```python
-import pandas as pd
-from time import perf_counter
-from c50py import C5Classifier
-
-df = pd.read_csv("titanic.csv")
-t0 = perf_counter(); clf.fit(X, y); print(f"fit: {perf_counter()-t0:.3f}s")
-
-# Inspect the tree
-clf.print_tree(feature_names=features, class_names=["No", "Yes"])
-
-# Extract rules for each sample
-rules = clf.predict_rule(X, feature_names=features)
-print(rules[:5])
-
-# Export as Graphviz
-path = clf.export_graphviz(
-    "titanic_tree",
-    feature_names=features,
-    class_names=["No", "Yes"],
-    format="dot"  # save a .dot file directly
-)
-print(f"DOT file written to {path}")
-```
-
-## Quickstart – Regression (Diabetes)
-
-Fit a regression tree to the diabetes dataset and obtain a visualisation:
-
-```python
-import pandas as pd
-from time import perf_counter
-from c5py import C5Regressor
-
-df = pd.read_csv("diabetes.csv")
-y = df["target"].values
-X_df = df.drop(columns=["target"])
-X = X_df.values.astype(object)
-features = list(X_df.columns)
-
-reg = C5Regressor(
-    min_samples_split=30,
-    min_samples_leaf=10,
-    pruning=True, cf=0.25, global_pruning=True,
-    feature_names=features,
-    random_state=42,
-    infer_categorical=False, int_as_categorical=False,
-    numeric_threshold_strategy="quantile", max_numeric_thresholds=64
-)
-
-start = perf_counter(); reg.fit(X, y); print(f"fit: {perf_counter()-start:.3f}s")
-
-# Export to DOT (Graphviz installed optional)
-dot_path = reg.export_graphviz("diabetes_tree", feature_names=features, format="dot")
-print(f"Tree saved to {dot_path}")
-
-# Export human‑readable rules (single trees only)
-rules = reg.export_rules(feature_names=features)
-print(rules[:3])
-```
-
-## Performance tuning
-
-Several hyperparameters influence model complexity and performance:
-
-- **`numeric_threshold_strategy`** (`'quantile'` | `'all'`): subsample candidate numeric
-  thresholds.  With `'quantile'` the number of splits considered is limited to
-  `max_numeric_thresholds` per feature per node.  `'all'` evaluates every unique
-  midpoint (slower on large datasets).
-- **`max_numeric_thresholds`**: number of candidate thresholds when using
-  `'quantile'` (typically 32–64).
-- **`categorical_features`**: list of names or indices marking categorical columns.
-- **`max_categories_exhaustive`**: maximum cardinality for exhaustive subset search on
-  categorical features; beyond this a simpler one‑vs‑rest strategy is used.
-- **`infer_categorical`/`int_as_categorical`**: enable automatic detection of
-  categorical/boolean/integer columns when dtype information is not explicit.
-- **`max_depth`**: optional depth limit for extremely noisy or deep trees.
-
-When boosting (`trials > 1`) the same hyperparameters apply to each base tree.
+# c50py — C5.0‑style Decision Trees for Python (clean 0.2.0)
+
+`c50py` provides transparent, easily inspectable decision trees modelled on
+Quinlan’s C5.0 algorithm.  Both classification and regression trees are
+supported and expose a scikit‑learn‑like API.  The implementation is written
+from scratch in pure Python/Numpy and includes support for numeric and
+categorical variables, missing values, pre‑ and post‑pruning, boosting,
+rule tracing/export and Graphviz visualisation.
+
+## Features
+
+- **Scikit-learn API:** `fit(X, y)`, `predict(X)`, `score(X, y)`.
+- **Categorical support:** Pass `categorical_features=[0, 2]` to handle categories natively.
+- **Sample weights:** Supports `sample_weight` in `fit` for weighted splitting and pruning.
+- **Missing values:** Handles missing values using C5.0's fractional propagation strategy.
+- **Boosting:** Set `trials=10` to train a boosted ensemble.
+- **Rule export:** call `export_rules()` to get a list of human-readable rules.
+- **Graphviz export:** call `export_graphviz()` to visualize the tree.
+- **Pretty printing:** call `print_tree` to display the learned splits in a
+  readable nested `if`/`else` format (single trees only).
+
+## Documentation
+
+For a comprehensive guide on how to use `c50py`, including advanced features and examples, please see the [Usage Guide](USAGE_GUIDE.md).
+
+## Installation (development mode)
+
+Install the package into your environment in editable mode:
+
+```bash
+pip install -e .
+```
+
+## Quickstart (Classification)
+
+```python
+import pandas as pd
+from time import perf_counter
+from c50py import C5Classifier
+
+df = pd.read_csv("titanic.csv")
+t0 = perf_counter(); clf.fit(X, y); print(f"fit: {perf_counter()-t0:.3f}s")
+
+# Inspect the tree
+clf.print_tree(feature_names=features, class_names=["No", "Yes"])
+
+# Extract rules for each sample
+rules = clf.predict_rule(X, feature_names=features)
+print(rules[:5])
+
+# Export as Graphviz
+path = clf.export_graphviz(
+    "titanic_tree",
+    feature_names=features,
+    class_names=["No", "Yes"],
+    format="dot"  # save a .dot file directly
+)
+print(f"DOT file written to {path}")
+```
+
+## Quickstart – Regression (Diabetes)
+
+Fit a regression tree to the diabetes dataset and obtain a visualisation:
+
+```python
+import pandas as pd
+from time import perf_counter
+from c50py import C5Regressor
+
+df = pd.read_csv("diabetes.csv")
+y = df["target"].values
+X_df = df.drop(columns=["target"])
+X = X_df.values.astype(object)
+features = list(X_df.columns)
+
+reg = C5Regressor(
+    min_samples_split=30,
+    min_samples_leaf=10,
+    pruning=True, cf=0.25, global_pruning=True,
+    feature_names=features,
+    random_state=42,
+    infer_categorical=False, int_as_categorical=False,
+    numeric_threshold_strategy="quantile", max_numeric_thresholds=64
+)
+
+start = perf_counter(); reg.fit(X, y); print(f"fit: {perf_counter()-start:.3f}s")
+
+# Export to DOT (Graphviz installed optional)
+dot_path = reg.export_graphviz("diabetes_tree", feature_names=features, format="dot")
+print(f"Tree saved to {dot_path}")
+
+# Export human‑readable rules (single trees only)
+rules = reg.export_rules(feature_names=features)
+print(rules[:3])
+```
+
+## Performance tuning
+
+Several hyperparameters influence model complexity and performance:
+
+- **`numeric_threshold_strategy`** (`'quantile'` | `'all'`): subsample candidate numeric
+  thresholds.  With `'quantile'` the number of splits considered is limited to
+  `max_numeric_thresholds` per feature per node.  `'all'` evaluates every unique
+  midpoint (slower on large datasets).
+- **`max_numeric_thresholds`**: number of candidate thresholds when using
+  `'quantile'` (typically 32–64).
+- **`categorical_features`**: list of names or indices marking categorical columns.
+- **`max_categories_exhaustive`**: maximum cardinality for exhaustive subset search on
+  categorical features; beyond this a simpler one‑vs‑rest strategy is used.
+- **`infer_categorical`/`int_as_categorical`**: enable automatic detection of
+  categorical/boolean/integer columns when dtype information is not explicit.
+- **`max_depth`**: optional depth limit for extremely noisy or deep trees.
+
+When boosting (`trials > 1`) the same hyperparameters apply to each base tree.

{c50py-0.2.2 → c50py-0.2.4}/pyproject.toml

@@ -1,37 +1,37 @@
-[build-system]
-requires = ["setuptools>=64", "wheel"]
-build-backend = "setuptools.build_meta"
-
-[project]
-name = "c50py"
-version = "0.2.2"
-description = "C5.0-like Decision Trees in Python (scikit-learn style)"
-readme = "README.md"
-license = { text = "MIT" }
-authors = [{ name = "David Díaz", email = "daviddiazsolis@gmail.com" }]
-requires-python = ">=3.9"
-dependencies = [
-    "numpy>=1.21",
-    "scikit-learn>=1.2"
-]
-
-classifiers = [
-    "Programming Language :: Python :: 3",
-    "License :: OSI Approved :: MIT License",
-    "Operating System :: OS Independent",
-    "Intended Audience :: Science/Research",
-    "Topic :: Scientific/Engineering :: Artificial Intelligence"
-]
-
-[project.optional-dependencies]
-graphviz = ["graphviz>=0.20"]
-
-[project.urls]
-Homepage = "https://github.com/daviddiazsolis/c50py"
-Issues = "https://github.com/daviddiazsolis/c50py/issues"
-
-[tool.setuptools.packages.find]
-where = ["src"]
-
-[tool.setuptools]
-include-package-data = true
+[build-system]
+requires = ["setuptools>=64", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "c50py"
+version = "0.2.4"
+description = "C5.0-like Decision Trees in Python (scikit-learn style)"
+readme = "README.md"
+license = { text = "MIT" }
+authors = [{ name = "David Díaz", email = "daviddiazsolis@gmail.com" }]
+requires-python = ">=3.9"
+dependencies = [
+    "numpy>=1.21",
+    "scikit-learn>=1.2"
+]
+
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "License :: OSI Approved :: MIT License",
+    "Operating System :: OS Independent",
+    "Intended Audience :: Science/Research",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence"
+]
+
+[project.optional-dependencies]
+graphviz = ["graphviz>=0.20"]
+
+[project.urls]
+Homepage = "https://github.com/daviddiazsolis/c50py"
+Issues = "https://github.com/daviddiazsolis/c50py/issues"
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.setuptools]
+include-package-data = true